Magnetic exploration beneath a near-surface magnetic noise source

Abstract

Exploration beneath a near-surface white magnetic noise source, such as maghemitic and lateritic soils or thin basalt flows, can best be achieved at ground level. The objective of such exploration is to discriminate between the noise and signal from underlying sources. The greater the ratio of signal-source depth to noise-source depth, the easier it is to distinguish between the two by filtering. Spectral analysis confirmed that low-pass linear filtering by either upward continuation (as implemented in airborne surveying) or digital processing (e.g. by running average or Butterworth filtering) is not appropriate because it does not distinguish between the low-frequency component of the noise originating near the surface and the low-frequency signal arising from a deep source. A filter was required which would cut all frequencies arising from random dipolar magnetic sources occurring between the surface and a selected depth. A running median filter with a window width greater than twice the expected noise-source depth (but less than twice the expected signal-source depth) has been confirmed by spectral analysis to attenuate the full spectrum of magnetic noise arising from dipolar sources within the surface noise-source layer. Application of the filter to fully sampled, ground-level data, recorded where surface maghemite overlies the Elura base-metal orebody near Cobar (NSW), demonstrated its effectiveness. The median filter may be further refined by defining a "range" value, judiciously selected such that the magnetic texture due to near-surface geological structure is preserved while still adequately discriminating against the low-frequency component of intense, random dipole noise. A typical range value selection is half the RMS noise amplitude. Signal-to-noise ratios were determined from the areas under the signal and filtered noise spectra. Graphing the signal-to-noise ratio against sensor elevation above the noise source demonstrated that the ratio achievable 0.5 m above ground level was five times better than that at the optimum airborne elevation of 75 m. Moreover, a survey elevation of 10 m (as may be conducted from a helicopter) would result in the very worst signal-to-noise ratio.